Utilizing a Spatial Grid for Automated Parking Space Vacancy Detection

In primarily populated areas, locating available parking can take time and effort, posing an environmental problem. Depending on the time of day, many drivers may face this issue due to the high volume of other drivers attempting to find parking. In most instances, urban drivers are forced to search for available street parking or vacant spaces in public parking lots. This behavior can often lead to traffic jams in local areas, commonly seen in populated cities such as Los Angeles and New York. Solutions to this public parking problem have introduced the idea of smart parking systems. In this research, we propose a novel approach to video-based parking space detection by utilizing a spatial grid to introduce localization to the scene. Our approach essentially utilizes a spatial grid that serves as a map of the scene, including only the road as cells within the grid. Once the grid is established, it encompasses the entirety of the parking lot, allowing our approach to use a network specialized in Monocular 3D Object Detection to map each vehicle's location more accurately within the scene with respect to the available parking spots identified during grid generation. To leverage the use of our system, we also built a demo application using a database to record the status of each parking space. By leveraging the duration of occupancy of each space, our system also has access to historical occupancy data, which can be used in tandem with other factors, like time of day and day of the week, to provide more valuable predictions and information that can assist drivers in finding parking more efficiently. We plan to continue thoroughly evaluating the accuracy and reliability of our system's predictions through comprehensive testing and comparisons with existing state-of-the-art systems to ensure its practicality in real-world scenarios